Color filters suitable for use in color image sensors have been described in U.S. Pat. Nos. 6,171,885, 6,395,576, 6,274,917, 6,482,669 and 6,495,813, all of which are incorporated by reference herein in their entireties.
A color image sensor is typically a charge-coupled device (CCD) or complementary metal-oxide semiconductor (CMOS) photodiode array structure. The structure includes a spectrally photosensitive layer below one or more layers patterned in an array of color filters, above which resides a surface-layer array of microlens elements. In some conventional configurations, a color pixel is formed using four adjacent pixels on an image sensor. Each of the four pixels is covered by a different color filter selected from the group of red, blue and two green pixels, thereby exposing each monochromatic pixel to only one of the three basic colors. Simple algorithms are subsequently applied to merge the inputs from the three monochromatic pixels to form one full color pixel.
FIGS. 1A and 1B show color filters having the “stripe defect.” The stripe defect occurs when one of the color resist layers is thicker or thinner than the nominal value. In FIG. 1A, the stripe defect is due to a difference between the thicknesses of the blue and green resist. The sensor 100 includes a plurality of scribe lines 120 arranged in a rectangular array. A plurality of filter regions are surrounded by the scribe lines 120. The filter regions include active regions 110 and bond pad regions 130 arranged within the scribe lines 120. The stripe defect is indicated by the plurality of regions, designated “B”, in which either the blue resist layer is thicker than its nominal value, or the green resist layer is thinner than its nominal value, and a plurality of regions designated “G” in which either the green resist layer is thicker than its nominal value, or the blue resist layer is thinner than its nominal value. Similarly, in FIG. 1B, the defect is due to a difference between the thicknesses of the green and red resist. In FIG. 1B, active regions are 110′, bond pad regions are 130′ and scribe lines are 120′. In regions designated “G”, either the green resist layer is thicker than its nominal value, or the red resist layer is thinner than its nominal value. In regions designated “R”, either the red resist layer is thicker than its nominal value, or the green resist layer is thinner than its nominal value.
FIG. 1D is a plan view of a bond pad exhibiting staining, which has also been observed when the bond pad is exposed to the color filter developing process. Such staining may indicate the presence of silicon oxide, organic residues, sodium, aluminium, fluorine and/or chlorine. Such staining is also associated with yield problems.
FIG. 1C is a side cross sectional view of a conventional CMOS imaging sensor (CIS) 100. The CIS 100 has a substrate 107 having a plurality of scribe lines 120 arranged to form at least one filter region 101 surrounded by the scribe lines. The substrate 107 may be an insulating substrate, or an insulating layer (e.g., SiO2) above a semiconductor substrate. In a wafer having many filter regions 101, the plurality of scribe lines 120 form a grid of perpendicular lines. The filter region 101 includes an active region 102 and a bond pad region 108 with bond pads 122 thereon. The active region 102 may have n− or p+ well photo diodes therein, and may be, for example polycrystalline silicon. Active region 102 has a passivation layer 114 (e.g., SiN) thereon. The passivation layer 114 has a plurality of top metal regions 103. The passivation layer 114 has a planarization layer 104a formed thereon, which may be a resist layer thicker than the height of the steps on the wafer surface. Planarization layer 104a can be cast on the wafer and can be planarized by, for example, a free viscous flow of the resist at an elevated baking temperature or by a forced resist flow through pressing a flat mold on the resist surface while heating. Above the first planarization layer 104a, blue, green and red color resist layers are formed, designated 111 and 113. A second planar layer 104b is above the filters 111 and 113. A microlens layer 106 is formed above the second planar layer 104b. 
The height 119 of layer 104b relative to the top of the passivation layer 114 may be about 4.5 μm. There is an additional height 118 of about 1.5 μm to 2 μm between the top of bond pad 122 and the top of the passivation layer 114, so the total step height between the top of passivation layer 114 and the top of the bond pad 122 is about 6 μm to 6.5 μm.
If the color filter signal deviation is too large, CMOS image sensor performance may be affected, resulting in yield loss as high as 15 to 20%. For example, the inventors of this application are aware of product samples in which the observed color signal standard deviations were: blue 3.93%, green 2.84% and red 1.68%. This corresponds to a yield measure Cpk of lower than 0.5.
The color filter deposition process and its relationship to the microlens array formation process are known to influence the production cycle-time, test-time, yield, and ultimate manufacturing cost.
A method of making a color filter with less signal deviation and improved yield is desired.